Multicast high availability enhancements for non-stop forwarding
View Patent ↗A distributed-forwarding router platform contains a master and a standby route processing manager (RPM). The master RPM uses dynamic internal routing codes to facilitate the replication of multicast packets within the router. As internal routing codes are assigned, the assignments are shared with the standby RPM. Should the standby RPM have to take over as the master RPM, the new master consults the internal routing codes assigned by the previous master as the new master builds multicast state, insuring that the internal routing codes the new master assigns are consistent with those used by the prior master. This allows the multicast forwarding plane to remain available during RPM failover, without a shutdown or unstable period while the new master RPM propagates internal routing codes. Other embodiments are also described and claimed.
1. A router, comprising:
a backplane;
a standby route processor management card coupled to the backplane; and
a master route processor management card coupled to the backplane, wherein the master route processor management card is operable to configure internal multicast data paths through the router, each internal multicast data path having at least one associated internal routing code corresponding to the internal multicast data path and at least one associated multicast group address; and
wherein the master route processor management card is also operable to transmit internal routing code updates to the standby route processor management card, the internal routing code updates informing the standby route processor management card as to associations between internal multicast data paths, internal routing codes, and multicast group addresses.
2. The router of claim 1 , wherein at least some of the internal routing codes comprise a multicast group identifier used to tag multicast data packets for internal transmission within the router, the multicast group identifier associated with a list of line cards in the router that are to receive a copy of a data packet internally tagged with the multicast group identifier.
3. The router of claim 1 , wherein at least some of the internal routing codes comprise an egress processing index used to tag multicast data packets for internal transmission within the router, the egress processing index associated, for at least one line card in the first router, with a list of ports on that line card that receive a copy of a data packet internally tagged with the egress processing index.
4. A method for switching between a master route processor manager and a standby route processor manager, comprising:
maintaining, on the master route processor manager, multicast state information for multicast group traffic paths traversing at least one port of a first router on which the master route processor manager operates;
maintaining a first internal routing code table comprising at least one internal routing code for each multicast group address for which data packets are duplicated in the first router, the internal routing code, for each such multicast group address, associated with the multicast group traffic paths that multicast data packets traverse through the first router;
copying, to a second internal routing code table local on a standby route processor manager, the internal routing codes stored in the first routing code table, and an association between the internal routing codes, the multicast group addresses served by the internal routing codes, and the traffic paths that data packets traverse through the first router for the internal routing codes; and
upon cessation of the operation of the master route processor manager as a multicast speaker, operating the standby route processor manager as the multicast speaker for the first router, the standby route processor manager using multicast neighbor router communications to construct multicast state information for multicast group traffic paths traversing at least one port of the first router, and the standby route processor manager reusing, for multicast group addresses saved in the second internal routing code table, the same internal routing codes associated with those multicast group addresses in the first internal routing code table.
5. The method of claim 4 , wherein the internal routing codes copied to the second internal routing code table for a given multicast group address are further associated with a specific multicast source.
6. The method of claim 4 , wherein the internal routing codes copied to the second internal routing code table for a given multicast group comprise a multicast group identifier, the multicast group identifier associated with a list of line cards in the first router that are to receive a copy of a data packet tagged with the multicast group identifier.
7. The method of claim 4 , wherein the internal routing codes copied to the second internal routing code table for a given multicast group comprises an egress processing index, the egress processing index associated, for at least one line card in the first router, with a list of ports that receive a copy of a data packet tagged with the egress processing index.
8. The method of claim 7 , wherein the list of ports further specifies, for at least one port in the list, a virtual local area network identifier to be inserted in a copy of the data packet to be forwarded out the at least one port.
9. The method of claim 8 , further comprising, when an egress processing index assigned by the first route processor management card is reused by the second route processor management card, transmitting the egress processing index and associated port information to a processor local to a group of egress ports, wherein the transmitting is delayed for a time after the second route processor management card takes over for the first route processor module.
10. The method of claim 4 , wherein at the end of an initial configuration period after the second route processor management card takes over for the first route processor management card, the second route processing management card removes internal routing code entries from the second internal routing code table that were copied from first routing code table but do not correspond to multicast state information constructed by the second route processor management card.
11. The method of claim 10 , wherein the initial configuration period ends upon the expiry of a tinier set to allow a set configuration time.
12. The method of claim 10 , wherein the initial configuration period ends after the at least one multicast neighbor router indicates that it has communicated all current multicast information to the second route processor management card.
13. The method of claim 4 , further comprising, when the second route processor management card receives, during an initial configuration period, multicast state information for a new multicast group traffic path that does not correspond to any multicast group address copied to the second internal routing code table from the first internal routing code table, creating a new multicast group traffic path internal routing code, for the new multicast group traffic path, that does not reuse an internal routing code copied from the first internal routing code table.
14. The method of claim 13 , further comprising installing the new internal routing code in a data forwarding path of the first router without waiting for the end of the initial configuration period.